Fix race condition in XNNPACK weights cache during concurrent init()

Differential Revision: D105753995

Pull Request resolved: pytorch#19682

Author: billmguo

backends/xnnpack/runtime/XNNPACKBackend.cpp

@@ -91,8 +91,13 @@ class XnnpackBackend final
     auto workspace = workspace_result.get();
 
     bool use_weight_cache = options_.resolve_weight_cache(context);
+    // Hold the lock for the entire init-compile-finalize sequence to prevent
+    // concurrent inits from resetting is_finalized_ or overwriting
+    // named_data_map_ while compileModel is using the shared weights cache.
+    std::unique_lock<std::mutex> lock_weights_cache(
+        weights_cache_mutex_, std::defer_lock);
     if (use_weight_cache) {
-      const std::lock_guard<std::mutex> lock_weight_cache(weights_cache_mutex_);
+      lock_weights_cache.lock();
       weights_cache_->initialize_for_runtime(
           context.get_runtime_allocator(), named_data_map);
     }